Upgrade to Pro — share decks privately, control downloads, hide ads and more …

The Human Processing Unit (HPU) as a New Approx...

Avatar for Yuhao Zhu Yuhao Zhu
March 24, 2016
Education
0
320

The Human Processing Unit (HPU) as a New Approximate Computing Substrate

WAX 2016 co-located with ASPLOS 2016

Avatar for Yuhao Zhu

Yuhao Zhu

March 24, 2016
Tweet

More Decks by Yuhao Zhu

See All by Yuhao Zhu
Automatically Generating Robotics Accelerators
Avatar for Yuhao Zhu queeten
0
62
RTNN: Accelerating Neighbor Search Using Hardware Ray Tracing
Avatar for Yuhao Zhu queeten
0
440
Architecture Support for Robust Deep Learning: Exploiting Software 1.0 Techniques to Defend Software 2.0
Avatar for Yuhao Zhu queeten
0
340
Watt-Wise Web: Architecting a Responsive and Energy-Efficient Mobile Web
Avatar for Yuhao Zhu queeten
0
94
Tail Latency in Node.js: Energy Efficient Turbo Boosting for Long Latency Requests in Event-Driven Web Services
Avatar for Yuhao Zhu queeten
0
790
Algorithm-SoC Co-Design for Mobile Continuous Vision
Avatar for Yuhao Zhu queeten
0
480
Energy and Power Measurement on Mobile Devices
Avatar for Yuhao Zhu queeten
0
180
Defense talk: Energy-Efficient Mobile Web Computing
Avatar for Yuhao Zhu queeten
0
640
GreenWeb: Language Extensions for Energy-Efficient Mobile Web Computing
Avatar for Yuhao Zhu queeten
0
400

Other Decks in Education

See All in Education
KBS新事業創造体験2025_科目説明会
Avatar for Yasuchika Wakayama yasuchikawakayama
0
160
1202
Avatar for cbtlibrary cbtlibrary
0
200
【ZEPホスト用メタバース校舎操作ガイド】
Avatar for aini school小中等部 ainischool
0
160
1014
Avatar for cbtlibrary cbtlibrary
0
510
くまのココロンともぐらのロジ
Avatar for Frieve-A frievea
0
140
俺と地方勉強会 - KomeKaigi・地方勉強会への期待 -
Avatar for ふぁらお加藤 pharaohkj
1
1.6k
Google Gemini (Gem) の育成方法
Avatar for MIKIO KUBO mickey_kubo
2
1.1k
東大１年生にJulia教えてみた
Avatar for Yusuke Matsui matsui_528
7
11k
子どもが自立した学習者となるデジタルの活用について
Avatar for Naoki Kato naokikato
PRO
0
170
Semantic Web and Web 3.0 - Lecture 9 - Web Technologies (1019888BNR)
Avatar for Beat Signer signer
PRO
2
3.2k
学習指導要領と解説に基づく学習内容の構造化の試み / Course of study Commentary LOD JAET 2025
Avatar for Masao Takaku masao
0
110
Use Cases and Course Review - Lecture 8 - Human-Computer Interaction (1023841ANR)
Avatar for Beat Signer signer
PRO
0
1.3k

Featured

See All Featured
Avoiding the “Bad Training, Faster” Trap in the Age of AI
Avatar for Mike Taylor tmiket
0
51
Exploring anti-patterns in Rails
Avatar for Elle Meredith aemeredith
2
230
Designing for Performance
Avatar for Lara Hogan lara
610
70k
Groundhog Day: Seeking Process in Gaming for Health
Avatar for Sebastian Deterding codingconduct
0
75
Rebuilding a faster, lazier Slack
Avatar for samanthasiow samanthasiow
85
9.3k
Navigating Team Friction
Avatar for Lara Hogan lara
191
16k
Build your cross-platform service in a week with App Engine
Avatar for Jose L jlugia
234
18k
AI: The stuff that nobody shows you
Avatar for John Nunemaker jnunemaker
PRO
2
170
実際に使うSQLの書き方 徹底解説 / pgcon21j-tutorial
Avatar for soudai sone soudai
PRO
196
71k
RailsConf 2023
Avatar for Aaron Patterson tenderlove
30
1.3k
Max Prin - Stacking Signals: How International SEO Comes Together (And Falls Apart)
Avatar for Tech SEO Connect techseoconnect
PRO
0
65
Leveraging Curiosity to Care for An Aging Population
Avatar for Cassini Nazir cassininazir
1
140

Transcript

  1. 1 WAX 2016 The Human Processing Unit (HPU) as a

    New Approximate Computing Substrate Yuhao Zhu UT Austin with Matt Lease, Matthew Halpern, and Vijay Janapa Reddi

  2. Computing Device

  3. Computing Device Problem Image Compression Edge Detection

  4. Computing Device Problem Image Compression Edge Detection Algorithm / Application

    JPEG 2000 Sobel

  5. Computing Device Problem Algorithm / Application HPU

  6. Computing Device Problem Algorithm / Application HPU What approximate problems

    can be mapped to HPUs?

  7. Computing Device Problem Algorithm / Application HPU What approximate problems

    can be mapped to HPUs? How do we program HPUs while ensuring quality?

  8. Computing Device Problem Algorithm / Application What approximate problems can

    be mapped to HPUs? How do we program HPUs while ensuring quality?